Follow the Oracle
The widespread availability of consumers' opinions shared on review platforms and other social media channels has offered a new source of information to better assess what products will succeed in the market place. Particularly, online reviews offer a source of information that is readily available and publicly accessible by providing a window to consumers' opinions as early as the time a product is launched. Online product reviews have been shown to be helpful in understanding and predicting product success in multiple contexts.
Even though this environment offers a very rich source of information, surprisingly, little to no work investigated reviewers themselves and whether their review behavior can help us to understand product success. The question is therefore whether the information on individuals that decide to write a review for a product early on after the product is launched can give companies, online platforms and other interested actors a signal for future product success.
We propose a novel approach to analyzing reviews data, by considering the reviewer, rather than the review itself, as the unit of analysis. We uncover that some reviewers have a predictive trait: they systematically review successful products early on after the product is launched. We use two large data sets of movies and books reviews and create for each reviewer in our data set a reviewer score, which measures the tendency of the reviewer to review successful products early on in the life cycle of the product. Using this approach, we can identify a segment of oracle reviewers who systematically provide early product reviews for products that end up being successful among consumers. We find not only that oracle reviewers exist, but they are predictive of product success over and beyond previously studied predictors of success in different domains. We can show that the higher the proportion of oracle reviewers in the crowd of product reviewers, the more successful the product. Oracle reviewers are predictive of various measures of product success, including sales, sales rank, product lifetime, and number of reviews.
Building on the phenomena of oracle reviewers, we uncover characteristics of oracle reviewers that systematically differentiate them from others. We find that oracle reviewers are selective reviewers, yet their predictive trait goes beyond selectivity. Oracle reviewers tend to write less reviews and to write them closer to the product's release date, relatively to non-oracle reviewers. Oracle reviewers do not like successful movies more than other reviewer and they write less emotional reviews, a characteristic that has been also found for critic reviewers.
Our work expands and combines the literatures of product reviews and predictive consumers by identifying a reviewer trait that makes them predictive of product success. We establish the existence of oracle reviewers across empirical contexts and platforms. Moreover, we show that predictive consumers can be identified even in the absence of observed purchase data and only based on their decision to review a product. In fact, we are not even using the content of the review to identify oracle reviewers, only whether they reviewed or not. In addition, the prevalence and availability of product reviews make our approach extremely useful in identifying predictive consumers. Marketers and platforms can use it to easily identify oracle reviewers who, by their mere choice to review a product, can inform them early on whether a product is going to be successful. This approach can be easily implemented by any company or platform whose products or services are reviewed by consumers. Online platforms such as Amazon can use this information to understand the success of products shortly after the product is launched and accordingly incorporate this information with respect to decisions what products to promote or decisions related to inventory. Advertisers can incorporate this information into how to allocate their budget and better place ads on products that will be popular. Manufacturers can monitor their products on review platforms to acquire an early understanding of what products are likely to be popular among consumers.